Classification of Thermal Breast Images Using Support Vector Machines

Abstract

Breast cancers are the most common types of cancer in the world and in our country. Breast cancer can significantly affect human health if not detected early. Today, the examination of mammography images by a physician, as well as physical intervention, is the traditional method used by the physician. This method contains x-rays that are harmful to human health and the preparation and shooting times of the shooting with this method can be long. Since breast cancers are more common in women, this also leads to the choice of a doctor. In this study, it has been studied on the classification of thermal breast images with the help of support vector machines in order to minimize such disadvantages caused by the methods used in the diagnosis of breast cancer. This method does not contain radiation and the processing time is very short compared to other methods. In the study, 30 thermal breast images were used, of which 15 images are patient images and the other 15 images are healthy individuals. Preliminary results with more than 70% accuracy were found by extracting features from the thermal images used. This situation is promising for the future goals of the study.